Proposal H, item 3: flat vertical scroll + sheet-level virtualization

[WilfordGrimley]

Description

Replaces /display's one-page-at-a-time pager with a continuous vertical stack of every sheet, per the owner's flat-scroll amendment. Each sheet's PagePreview mounts only when on/near screen via the existing RenderIfVisible component (already proven in CardResultSet.tsx) rather than a new IntersectionObserver implementation from scratch.

• sheets (derived from displayPagination.ts's paginateSlotsForDisplay) replaces the old single-page currentPageEntries/currentPageSlots derivation — every sheet's slot content is resolved (cheap bookkeeping), but only sheets within one sheet-height of the viewport actually mount real <img> tags.
• The toolbar's old prev/next pager (display-page-indicator) is gone. In its place: a passive display-sheet-indicator readout ("Sheet N of M") driven by its own IntersectionObserver — a thin band near vertical centre, distinct from RenderIfVisible's own broader mount/unmount rootMargin.
• handleSlotClick/selected-slot-index logic now operates across all sheets simultaneously, not just "the current page."
• This branch also carries a merge of origin/master to reconcile with item 2 (Proposal H, item 2: inline PDF export runs the real pipeline on /display #109, inline PDF export) and item (c) (Item (c): requested-printing badge on editor slots #110, RequestedPrintingBadge extraction), both of which landed on master after this branch was created and both touch DisplayPage.tsx significantly. The merge was hand-resolved (one import-line conflict) and re-verified in full post-merge — see the checklist below.

Benchmark (owner-required gate)

Measured against a 120-card deck (15 sheets) under 4x CPU throttle, dev-mode flag-on build, two runs:

metric	run 1	run 2
avg fps	58.7	59.6
p95 frame time	21.9ms	21.3ms
peak JS heap	236.8 MB	263.1 MB
max simultaneously-mounted <img> tags	16 of 120	16 of 120
long tasks (>50ms) during scroll	0	0

Decision: sheet-level virtualization is sufficient — shipping as-is, no row-granular fallback needed. Both runs meet the ~60fps threshold with zero long tasks and a bounded image-mount count (~2 sheets' worth regardless of deck size), confirming the anti-crash goal holds. Caveat: measured under Next dev mode (unminified, React dev overhead) via the project's own Playwright dev-server harness, not the literal built production bundle — a real production build would only improve on these numbers, so this is a conservative/pessimistic reading, not an optimistic one.

Follow-up (owner-requested): the benchmark script is now committed under frontend/tests/perf/display-scroll.bench.spec.ts, runnable via npx playwright test --config=playwright.perf.config.ts — a separate config so it's excluded from both a plain npx playwright test and CI's sharded run (playwright.config.ts now testIgnores tests/perf/**). Re-running it against this branch reproduces the original numbers (120 cards/15 sheets, ~59fps, peak heap ~257MB, 16/120 max mounted <img>s, 0 long tasks). Intended to give the upcoming pane migration (which restructures this exact surface) a number to compare against instead of a feeling.

Checklist

• I have installed pre-commit and installed the hooks with pre-commit install before creating any commits.
• I have updated any related tests for code I modified or added new tests where appropriate. New Playwright test: multi-sheet deck renders as a continuous stack with the far-off sheet deferred until scrolled into view (DisplayPage.spec.ts). Existing display-page-indicator assertion updated to display-sheet-indicator. Benchmark script added under tests/perf/ (non-CI, manually run).
• I have manually tested my changes as follows:
  • npx tsc --noEmit clean
  • npx jest — 42/42 suites, 399/399 tests passing
  • NEXT_PUBLIC_UNIFIED_DISPLAY_ENABLED=true npx next build — flag-on production build succeeds
  • npx playwright test tests/DisplayPage.spec.ts tests/DisplayPageExport.spec.ts — 19/19 passing on the merged tree (full-file runs, not filtered)
  • npx playwright test --list (default config) confirms tests/perf/ is invisible to normal/CI runs; npx playwright test --config=playwright.perf.config.ts runs it and reproduces the benchmark numbers
  • Manual benchmark run (see above), twice for stability
• I have updated any relevant documentation or created new documentation where appropriate. Item 3 didn't require a design-doc-lite update per the owner's "ASCII-mockup-in-PR-body suffices for structural PRs" waiver for this work order.

Known sandbox flakiness (documented per prior PRs in this sequence, not chased further): unfiltered full-file Playwright runs in this dev sandbox intermittently fail with "Playwright Test did not expect test.describe()/describe() to be called here," reproduced even on untouched spec files — confirmed environmental, not code-related. Not observed on the runs quoted above.